def test_load(self):
pubs, stats = gs.parse_dblp_page(open(self.test_file).read())
self.assertEqual(len(pubs), 31)
self.assertEqual(
pubs[0]['title'],
'Detecting Change Points in the Large-Scale Structure of Evolving Networks.'
)
self.assertEqual(pubs[0]['pub_type'], 'inproceedings')
def test_load(self):
pubs, stats = gs.parse_dblp_page(open(self.test_file).read())
self.assertEqual(len(pubs), 31)
self.assertEqual(pubs[0]["title"], "Detecting Change Points in the Large-Scale Structure of Evolving Networks.")
self.assertEqual(pubs[0]["pub_type"], "inproceedings")
gs_file = os.path.join(args.gs_dir, GS_FILE % (f['gs'], num_loaded))
all_pubs = []
stats = None
while os.path.isfile(gs_file):
pubs, stats = parse_gs_page(open(gs_file).read())
all_pubs += pubs
num_loaded += 1
gs_file = os.path.join(args.gs_dir, GS_FILE % (f['gs'], num_loaded))
output_file = os.path.join(args.gs_dir, GS_PKL % f['gs'])
with open(output_file,'wb') as fp:
pickle.dump(all_pubs, fp)
pickle.dump(stats, fp)
if 'dblp' in f and args.dblp_dir is not None:
dblp_file = os.path.join(args.dblp_dir, DBLP_FILE % (f['dblp'], 0))
all_pubs, stats = parse_dblp_page(open(dblp_file).read())
for pub in all_pubs:
role = get_author_role(f.facultyName, pub['authors'])
pub['author_role'] = role
output_file = os.path.join(args.dblp_dir, DBLP_PKL % f['dblp'])
with open(output_file,'wb') as fp:
pickle.dump(all_pubs, fp)
pickle.dump(stats, fp)
print num_processed, f['facultyName']
num_processed += 1